Forecasting forest fire burning area using machine training

Abstract

The objective of this article is to create and train an artificial neural network based on a data set containing various climatic parameters and future fire area as an output parameter that the authors intend to predict. Such a “set” of data is usually available for research and study. Before training the neural network model, the data set is divided into two samples: a sample for training, which is about 90% of the set; and a sample for testing the trained model. In setting the task, the authors select and analyze the known data on the fires that occurred in Montesinho Park, compare the models trained on these data with and without normalization. As a result, two examples are given of a qualitative demonstration of graphs of absolute error changes of fire areas, which are projected using the created and trained model.

Keywords

• площадь горения
• машинное обучение
• модель
• нейронные сети
• keras
• прогнозирование
• лесной пожар